Foundations of Generative AI

Georgia Institute of Technology via edX

Go to class Write review

Details

Go to class

Provider

edX
Pricing

Free Online Course (Audit)
Languages

English
Certificate

$90.00 Certificate Available
Duration & workload

3 weeks, 2-3 hours a week
Sessions

On-Demand
Subtitles

English

Found in

Overview

This course provides an accessible but comprehensive introduction to the world of generative artificial intelligence. Organized into ten modules, it takes you on a journey from the earliest ideas of artificial intelligence to the state‑of‑the‑art techniques that power today’s most advanced generative systems.

This innovative course was largely produced using generative AI tools — including AI‑generated/assisted video instruction, interactive exercises, and AI‑supported grading — and demonstrates responsible, real‑world use of these technologies in learning.

You’ll begin by learning the history of AI, exploring classical approaches based on symbolic reasoning, the rise of machine learning and statistical methods, and the recent breakthroughs that have enabled generative AI models to create text, images, and code. From there, you’ll explore what makes generative AI unique: its ability to produce new content by predicting patterns from massive datasets.

This artificial intelligence course will then guide you through the core mechanics of these systems. You’ll learn how models break down text into tokens, how they use context windows to process information, and how neural networks — evolving from simple feedforward models to recurrent, convolutional, and ultimately transformer architectures — enable machines to generate coherent, creative outputs. You’ll examine attention mechanisms, training methods, and fine‑tuning techniques, as well as the alignment strategies used to ensure these systems behave in ways consistent with human goals and values.

Throughout, the generative AI course emphasizes not just theory but application. You’ll see how generative AI is used in education, business, software development, and creative fields, and you’ll gain the ability to analyze both its strengths and its limitations. In keeping with its theme, this course itself is innovative: much of the instruction, practice, and even grading is generated or assisted by AI tools, making it one of the first educational experiences to showcase generative AI not just as a subject of study, but as a partner in teaching.

By the end of the course, you’ll have a clear understanding of how generative AI works, why it matters, and how to integrate it responsibly into your own work. Whether you are a professional, student, or simply curious about AI, you’ll be equipped with both foundational knowledge and practical insights to navigate this transformative technology.

Syllabus

Module 0: Prologue and Analogy
Introduction to the course, use of AI avatars for instruction, and the automobile analogy that frames user roles in generative AI adoption.

Module 1: A Brief History of AI
Survey of classical AI (rule-based systems), statistical AI (machine learning and pattern recognition), and generative AI (creative prediction).

Module 2: What Is Generative AI?
Core ideas behind generative AI: pattern completion, probabilistic creativity, and generative generalization, with examples across text, images, and code.

Module 3: Why Generative AI Matters
Practical significance of generative AI: individual empowerment, new paradigms of human-computer interaction, and integration into global technology.

Module 4: From Patterns to Predictions
How generative AI models process data: tokens, context windows, probability distributions, and next-token prediction.

Module 5: Neural Networks and Their Evolution
Overview of neural networks as function approximators, including feedforward, recurrent, and convolutional architectures.

Module 6: Transformers and Generative Output
The breakthrough of transformer architectures, self-attention, positional encoding, and decoding strategies like greedy, sampling, and top-k and top-p methods.

Module 7: Training Generative Models
How models are trained at scale: stochastic gradient descent, loss functions, data requirements, and the role of computation.

Module 8: Fine-Tuning Models
Adapting pretrained models for specific domains or styles through fine-tuning and parameter-efficient methods.

Module 9: Aligning Generative AI
Ensuring AI systems behave responsibly: goal, behavior, and value alignment; reinforcement learning from human feedback (RLHF); constitutional AI; and ongoing ethical challenges.